On Monday, Bilal Chughtai, a former DeepMind research engineer, posted on X that he had resigned in July 2026 and that he feared AI could pose an existential threat.
CHUGTAI joined a growing list of experts who are raising alarms. Last week, Anthropic senior researcher Jacob Coxon resigned, citing a "greater than 10% chance" that AI could end humanity by the decade's end, a claim echoed by fellow scientist Evan Hubinger.
While at DeepMind, Chughtai co‑authored papers on alignment and AGI safety, most notably a 2024 study on value alignment that warned of runaway incentives. His work has been cited in several internal safety reviews.
He urged the industry to slow progress, warning of a "maniac race" that could outpace society's ability to manage new risks. "We need to pace AI development to a speed that society can handle," he said.
Google declined to comment outside business hours, but regulators in the U.S. and EU are drafting a framework for joint safety audits, with a summit slated for October 2026.
In a lab near the campus, a junior engineer named Neha recalled the late‑night code sprints, a reminder that the human cost of unchecked AI could be personal.